Course in Artificial Intelligence for Insurance Fraud
Online 300-hour course in artificial intelligence for insurance fraud. 6 modules covering Regulation (EU) 2024/1689 on artificial intelligence, Regulation (EU) 2016/679 (GDPR) and Python (scikit-learn, XGBoost). Assigned tutor and enrolment open.
Take the controls.
Run a real engineering trade-off from this field. Move the sliders, watch the model react in real time and beat the challenge.
What you will be able to do
Frame insurance fraud as a data problem.
Prepare variables from claims, policies and relationships.
Train supervised and unsupervised detection models.
Evaluate models using metrics suited to imbalanced classes.
Integrate alerts into the investigation process.
Ensure explainability, fairness and regulatory compliance.
Deploy and monitor an anti-fraud model in production.
Who it's for
- Anti-fraud data scientist looking to specialise in artificial intelligence in insurance fraud.
- Anti-fraud unit manager looking to specialise in artificial intelligence in insurance fraud.
- Machine learning engineer looking to specialise in artificial intelligence in insurance fraud.
- Claims intelligence analyst looking to specialise in artificial intelligence in insurance fraud.
- Model auditor looking to specialise in artificial intelligence in insurance fraud.
- Anti-fraud solutions consultant looking to specialise in artificial intelligence in insurance fraud.
- Graduates in engineering or sciences seeking to join insurance companies.
What you'll work with
- Regulation (EU) 2024/1689 on artificial intelligence
- Regulation (EU) 2016/679 (GDPR)
- Organic Law 3/2018 (LOPDGDD)
- ISO/IEC 42001 (artificial intelligence management systems)
- ISO/IEC TR 24027 (bias in AI systems)
- Criminal Code, art. 248 (fraud)
- Law 50/1980 on Insurance Contracts
6 modules · 300 h
1.1Typologies by line of business: motor, home, health and sick leave1.2Opportunistic fraud versus organised fraud1.3Application fraud and claims fraud1.4Available data: claim file, history, network and external sources1.5Labels: confirmed, suspected and undetected fraud1.6Label bias caused by the prior detection process1.7Real volume versus detected volume1.8Cost of fraud and cost of detection1.9What artificial intelligence can and cannot contribute1.10Workshop: data and label map of an insurer
2.1Building the analysis table per claim2.2Variables from the claim, the customer and history2.3Time variables: policy age and proximity of events2.4Network variables: repeated parties and repair shops2.5Free text in the claim file and its use2.6Image data: damage photographs2.7External sources and common industry files2.8Information leakage and contaminated variables2.9Documentation of each variable and its origin2.10Practice: building a set of variables
3.1Supervised models on confirmed cases3.2Extreme class imbalance: techniques and risks3.3Gradient boosting and regularised linear models3.4Anomaly detection without labels3.5Clustering to discover new patterns3.6Graph analysis for organised fraud3.7Natural language processing on the claim narrative3.8Computer vision in damage assessment3.9Combining models and rules in a decision engine3.10Practice: comparing three approaches on the same data
4.1Metrics: precision, recall, precision-recall curve4.2Cost of false positives and false negatives4.3Choosing the threshold based on cost4.4Real capacity of the investigation team4.5Prioritisation and work queue4.6Explaining the alert to the investigator4.7Integration with the claims system4.8Measuring the model's effect on outcomes4.9Fraud drift and retraining4.10Practice: choosing a threshold with a cost function
5.1Automated decisions and the right to human intervention5.2Local explanation of an alert and its quality5.3Sensitive variables and indirect proxies5.4Impact analysis by customer groups5.5Legal basis for processing and sector files5.6Customer information and complaints procedure5.7Traceability of each engine decision5.8European artificial intelligence regulation and its fit5.9System audit and independent review5.10Practice: explainability sheet for an anti-fraud model
6.1Model governance and owners of each stage6.2Production monitoring and degradation alerts6.3Feedback loop with investigators6.4Version control and computing environment6.5Training of claims teams6.6Industry collaboration and limits of data sharing6.7Measuring the programme's return6.8Common mistakes when deploying an anti-fraud model6.9Anti-fraud maturity roadmap6.10Final project: complete anti-fraud system with its governance
How you'll learn
Fully 100 % online training on the SEIUM campus, with downloadable material, recorded classes and an assigned technical tutor. No fixed timetables: students progress at their own pace within the intake, allowing it to be combined with work.
Each module combines theory, guided reading of the original standard and applied exercises. The working software is Python (scikit-learn, XGBoost), Neo4j, SAS Fraud Framework, Shift Technology, SHAP, in free, academic or evaluation versions. Trials are studied using real procedures of detection models with imbalanced data, collusion network analysis and threshold tuning according to investigation capacity. Assessment is continuous per module plus the final integrating project.
Build it end to end
Your final project follows the real engineering workflow, step by step, reviewed by your tutor.
Frame insurance fraud as a data problem.
Prepare variables from claims, policies and relationships.
Train supervised and unsupervised detection models.
Evaluate models using metrics suited to imbalanced classes.
Integrate alerts into the investigation process.
Where this takes you
Requirements
- No prior degree required.
- Internet connection and a computer capable of running the programme's software.
- Sufficient Spanish to follow technical regulatory reading.
Ready to start?
Send your application today. Our admissions team replies within 72 hours with your study plan and funding options.
Frequently asked questions
The programme comprises 300 total hours spread over 3 to 5 months. That is between 10 and 15 hours per week on average, compatible with full-time work.
Work is done with Python (scikit-learn, XGBoost), Neo4j, SAS Fraud Framework, Shift Technology, SHAP. Free, academic or evaluation versions are used, and the tutor indicates the free alternative where one exists.
An own diploma from SEIUM upon passing continuous assessment and the integrating project. It is in-house training, not an official Ministry qualification.
Yes. Modules passed in a course or postgraduate diploma in the same area are recognised in the higher programme. Check the recognition policy when enrolling.
Related programmes
Course in Innovative Business Models in Insurtech
Online 300-hour course in innovative business models in insurtech. 6 modules covering Directive (EU) 2016/97 (insurance distribution, IDD), Law 7/2020 (digital transformation of the financial system,…
Seismic Risk in Urban Infrastructure Course
300-hour online course in seismic risk in urban infrastructure. 6 modules covering EN 1998 (Eurocode 8, design of earthquake-resistant structures), NCSE-02 (Spanish Seismic Construction Standard) and…
Electric Vehicle and Battery Insurance Course
Online 300-hour course in electric vehicle and battery insurance. 6 modules covering Regulation (EU) 2023/1542 (batteries), UN Regulation No. 100 (electrical safety of electric vehicles) and…
More in Insurance, Risk & Actuarial Data
All Insurance, Risk & Actuarial Data programmes → · How to apply · Fees & financing · Scholarships